Updraft Model for Development of Autonomous Soaring Uninhabited Air Vehicles - NTRS

Page created by James Palmer
 
CONTINUE READING
https://ntrs.nasa.gov/search.jsp?R=20060004052 2020-02-19T23:47:43+00:00Z

      Updraft Model for Development of Autonomous Soaring
                    Uninhabited Air Vehicles

                                                Michael J. Allen *
                     NASA Dryden Flight Research Center, Edwards, California 93523-0273, USA

              Large birds and glider pilots commonly use updrafts caused by convection in the lower
          atmosphere to extend flight duration, increase cross-country speed, improve range, or
          simply to conserve energy. Uninhabited air vehicles may also have the ability to exploit
          updrafts to improve performance. An updraft model was developed at NASA Dryden Flight
          Research Center (Edwards, California) to investigate the use of convective lift for
          uninhabited air vehicles in desert regions. Balloon and surface measurements obtained at the
          National Oceanic and Atmospheric Administration Surface Radiation station (Desert Rock,
          Nevada) enabled the model development. The data were used to create a statistical
          representation of the convective velocity scale, w*, and the convective mixing-layer thickness,
          zi. These parameters were then used to determine updraft size, vertical velocity profile,
          spacing, and maximum height. This paper gives a complete description of the updraft model
          and its derivation. Computer code for running the model is also given in conjunction with a
          check case for model verification.

                                                   Nomenclature
Cp           =   specific heat of dry air, J/kgºK
es           =   saturated vapor pressure, mb
ev           =   vapor pressure, mb
g            =   gravitational constant, m/s2
h            =   height above ground, m
j            =   ground measurement sample index
ki           =   shape constant, unitless
L            =   straight line length, m
max          =   maximum
n            =   ground measurement sample index at sunset
N            =   number of updrafts inside a given area
NL           =   number of updrafts encountered along straight line
p            =   surface pressure, mb
p0           =   reference pressure, mb
PDF          =   probability density function
QG           =   heat into ground, W/m2
 Q! H        =   sensible heat flux, W/m2
QH           =   kinematic sensible heat flux, °K*m/s
QOV          =   surface virtual potential temperature flux, W/m2
QS           =   net radiation at surface, W/m2
r            =   distance to updraft center, m
r1           =   updraft core radius, m
r2           =   updraft outer radius, m
rh           =   relative humidity, percent
SURFRAD      =   Surface Radiation
swd          =   downdraft velocity ratio, unitless
T            =   temperature, °C

*
    Aerospace Engineer, Controls and Dynamics Branch, P.O. Box 273/MS-4840D, Member.

                                                          1
                                  American Institute of Aeronautics and Astronautics
Ts          =   surface temperature, °C
UAV         =   uninhabited air vehicle
w           =   updraft velocity, m/s
w*          =   convective velocity scale, m/s
w1          =   intermediate variable in downdraft velocity calculation, m/s
wc          =   updraft velocity corrected for environment sink, m/s
wD          =   downdraft velocity, m/s
we          =   environment sink velocity, m/s
wpeak       =   maximum vertical velocity, m/s
ws          =   mixing ratio, g/kg
w           =   average updraft velocity, m/s
X           =   test area length, m
Y           =   test area width, m
z           =   aircraft height, m
zi          =   convective mixing-layer thickness, m

β           =   Bowen ratio, unitless
ρ           =   moist air density, kg/m3
θ0          =   surface potential temperature, °K
σ           =   standard deviation
ΓD          =   dry adiabatic lapse rate, °C/m
ψ           =   angular position, rad
!0          = daily average surface potential temperature, °K

                                                I.   Introduction

O     ne relatively unexplored way to improve the range, duration, or cross-country speed of an autonomous aircraft
      is to use buoyant plumes of air found in the lower atmosphere, known as thermals or updrafts. Updrafts occur
when the air near the ground becomes less dense than the surrounding air because of heating or humidity changes at
the surface of the Earth. Piloted sailplanes rely solely on this free energy to make flights of more than
2000 kilometers.1 Soaring birds such as hawks and vultures have been observed to circle for hours without flapping
their wings; and Frigatebirds are known to soar continuously, day and night, using updrafts.2
    Many uninhabited air vehicles (UAVs) have similar sizes, wing loadings, and missions as soaring birds and
sailplanes. Such missions that could allow small UAVs to take advantage of updrafts encompass remote sensing,
surveillance, atmospheric research, communications, and Earth science. A previous study using a simple
UAV simulation with updrafts calculated from measured surface and balloon data found that a 2-hr nominal-
endurance UAV can potentially gain 12 hr of flight time during the summer and 5 hr of flight time during the winter
using updrafts.3
    Researchers have used various updraft models to study optimal soaring paths,4 glider design,5 and cloud
formation.6 Wharington used a simple updraft model to develop algorithms for a soaring UAV.7 These models
generally provide a vertical velocity distribution for a given radius and peak updraft velocity. This paper presents a
unique updraft model in that statistical values for updraft velocity, convective mixing-layer thickness, updraft
spacing, and updraft size are determined from measured data.

                                    II.   Convective-Layer Scale Factors
    The convective velocity scale, w*, and convective mixing-layer thickness, zi, were calculated from surface and
rawinsonde balloon measurements taken at the National Oceanic and Atmospheric Administration (NOAA) Surface
Radiation (SURFRAD) station in Desert Rock, Nevada (lat. 36.63 deg N, long. 116.02 deg W, elev. 1007 m). The
convective layer is the lowest region of the atmosphere where significant mixing occurs. During calm conditions,
buoyant plumes of air that have been heated at the surface cause local mixing. These plumes of air, called updrafts
or thermals, can have significant upward velocity.
    SURFRAD instrumentation includes a radiometer platform, meteorology tower, and solar tracker. Measurements
were taken every 3 min. A rawinsonde station is also collocated with the SURFRAD site where rawinsonde balloons

                                                         2
                                 American Institute of Aeronautics and Astronautics
were launched every 12 hr. This study used surface temperature, wind, and radiation measurements as well as
balloon-measured temperature and humidity collected from the entire year of 2002.8
    The convective mixing-layer thickness, zi, as Fig. 1 shows, is the maximum height-above-ground that updrafts
generally obtain. The mixing-layer thickness was calculated using predawn rawinsonde balloon data and measured
surface temperatures. The convective mixing-layer thickness was estimated by finding the intersection of the
predawn balloon-measured temperature profile as a function of altitude with the line given by Eq. (1).

Figure 1. Simplified representation of atmosphere with updrafts showing convective mixing-layer
thickness, zi.

                                                  T = " !D h + Ts                                                   (1)

   The line defined by Eq. (1) gives a temperature, T, for every height, z, using the dry adiabatic lapse rate, Γ D, of
0.00975 °C/m. Figure 2 shows an example of the zi calculation.

                                        Figure 2. Example of zi calculation.

    A surface heat budget was used to calculate the convective velocity scale, w*. The first step in this process is to
calculate the sensible heat flux, Q! H , given by Stull in Eq. (2).9

                                                         3
                                 American Institute of Aeronautics and Astronautics
~    !("Q S + QG )
                                                   QH =                                                                  (2)
                                                           (1 + !)

    In this equation, Q! H is the sensible heat flux, QS is the net radiation at the surface, QG is the heat into the ground,
and β is the Bowen ratio, defined as the ratio of sensible to latent heat fluxes at the surface. This study uses a Bowen
ratio of 5, given for semi-arid regions.9 The net radiation, Q S, was measured directly by the SURFRAD station; and
the heat into the ground, QG, was calculated applying Eq. (3) taken from Stull9 using the percentage method for
daytime calculations.

                                                      QG = 0.1QS                                                         (3)

   Sensible heat flux was then converted to its kinematic form using Eq. (4).

                                                               Q! H
                                                       QH =                                                              (4)
                                                               !C p

    Values for the density of moist air, ρ, and the specific heat of dry air, Cp, used in Eq. (4), were 1.210 kg/m3 and
1004.67 J/kgºK, respectively. The surface virtual potential temperature flux, QOV, was then calculated with Eq. (5)
using the definition of virtual potential temperature for unsaturated air.9

                                                QOV = Q H (1 + 0.61ws )                                                  (5)

   Equation (5) was solved using the mixing ratio, ws, calculated from Eq. (6).

                                                              622ev
                                                      ws =                                                               (6)
                                                              p ! ev

   In Eq. (6), p is the surface pressure and ev is the vapor pressure given by Eq. (7).

                                                      ev = rh•es/100                                                     (7)

   Equation (7) was solved using the measured relative humidity, rh, and saturated vapor pressure, es, calculated
from Eq. (8).

                                                                  17.67 Ts
                                                                  Ts + 243.5
                                                   es = 6.112e                                                           (8)

   Equation (8) was solved using measured surface temperature, Ts. Equations (2) to (8) determine the surface heat
budget required to calculate Q OV. Equation (9) calculates the convective scaling velocity.

                                                                       1
                                                       "            g %3
                                                  w* = $ Qov zi '                                                        (9)
                                                       #       !0 &

   In Eq. (9), g is the gravitational constant and     !0 is the daily average surface potential temperature given by
Eq. (10).

                                                          4
                                  American Institute of Aeronautics and Astronautics
0.286

                                                  (               )
                                              n            "p %
                                           ( Ts j + 273.15 $ p0 '
                                           j=1             # j&
                                      !0 =                                                                          (10)
                                                       n
    Equation (10) was solved using measured surface temperature, Ts, and pressure, p, and a reference pressure, p0,
of 1000 mb. The sample index, j = 1, refers to sunrise and j = n refers to sunset. The convective velocity scale, w*, is
used primarily to calculate the updraft vertical velocity. The convective scale velocity was set to zero if the surface
wind velocity was greater than 12.87 m/s (25 knots) to account for the disruptive effect of high winds on updrafts.

                                           III.       Test Point Matrix
    This study uses statistical tools to summarize yearly and daily changes in zi and w* into a set of representative
cases. The histogram in Fig. 3 shows the distribution of w* for all daylight hours during the year 2002. Data points
that produced a convective velocity scale of zero comprise 25 percent of the total set of recorded data and were not
included in the Fig. 3 histogram. Zero convective mixing-layer thickness, zero net radiation, or high winds can cause
zero scale velocity. Of these causes, the most prevalent was found to be zero net radiation because of low sun angles
during early morning and late evening.

   Figure 3. The w* histogram for all samples during daylight hours of 2002, when updrafts were present.

    A statistical representation of w* was found using percentiles. The following percentiles were chosen as test
points: 2.3, 15.9, 50.0, 84.1, and 97.7. The five test cases are referred to as –2σ, –1σ, mean, +1σ, and +2σ because
they are analogous to the standard deviation and mean values of a Gaussian probability density function (PDF). The
use of percentiles allowed the calculation of a simple set of test cases to represent the distribution of convective
velocity scale.
    A distribution of all zi values for each selected w* in the test point matrix was created by collecting all zi points
corresponding to w* values that fell within a window of ±0.1 m/s from the selected test point w*. Figure 4 shows an
example of the zi data selection for a given w*. A Gamma PDF was used to model the statistics of zi. The Gamma
PDF coefficients that best fit the data were used to calculate the –1σ, mean, and +1σ points in the test point matrix.
Table 1 gives the resulting set of convective scale factor test points.

                                                         5
                                 American Institute of Aeronautics and Astronautics
Figure 4. Convective mixing-layer thickness, zi, as function of convective velocity scale, w*. (Dashed lines
indicate zi window for the example case of w* = 4.08 m/s.)

                 Table 1. Convective scale test points during times when updrafts are present.

                         Description           w*, m/s      –1σ zi, m     mean zi, m     +1σ zi, m

                   –2σ w*                       0.46           25.6          53.6          97.4

                   –1σ w*                       1.27           150           210           1007

                   mean w*                      2.56           767           1401          2319

                   +1σ w*                       4.08          2134           2819          3638

                   +2σ w*                       5.02          2913           3647          4495

    The test conditions given in Table 1 indicate the year-round statistical properties of convective lift during times
when updrafts were present. These test points include variations in convection as a result of time of day and time of
year. Table 1 shows large variations of w* and zi, illustrating the challenge presented to a soaring aircraft that must
use as many updraft sizes and strengths as possible.
    Table 2 presents monthly trends. These points represent the mean and maximum values of w* during the daylight
hours of the selected month. The zi test points in Table 2 were found by taking the mean of all zi points in the
selected month that fell within a window of ±0.1 m/s from the selected test point w*. Test points given in Table 2
reveal the seasonal variations in updraft strength and altitude. Daily variations are generally sinusoidal, with zero w*
at sunrise, maximum w* near noon, and zero w* at sunset.

                                                         6
                                 American Institute of Aeronautics and Astronautics
Table 2. Monthly convective scale test points for all times between sunrise and sunset.

                           Jan    Feb     Mar    Apr    May      Jun    Jul    Aug     Sep    Oct    Nov     Dec

  mean w*, m/s             1.14   1.48    1.64   1.97   2.53     2.38   2.69   2.44    2.25   1.79   1.31   1.26

  zi for mean w*, m        504    666     851    1213   1887     1728   1975   1755    1382   893    627     441

  max w*, m/s              3.59   3.97    4.89   5.53   5.49     5.51   6.30   5.64    5.97   4.57   4.55   4.11

  zi for max w*, m        1800    1970   3900    2380   3833     4027   3962   4940    2460   3285   1783   1680

                                          IV.    Updraft Calculations
          *
    Each w and zi scaling parameter given in Table 1 and Table 2 can be used to calculate updraft velocity, radius,
and vertical velocity distribution using the equations provided in this section. Equation (11) taken from Lenschow10
calculates average updraft velocity using height, z, and the convective-layer scale parameters, w* and zi,
from Table 1 or Table 2.

                                                           1
                                                    *! z $ 3 ! z$
                                            w = w # & # 1 ' 1.1 &                                               (11)
                                                  " zi % "     zi %

   Figure 5 is a plot of Eq. (11). The maximum vertical velocity of 0.45 w* occurs at a height ratio, z/zi, of
approximately 0.25. Also note that the updraft velocity is negative for height ratios greater than 0.9. The height at
which peak vertical velocity occurs is lower than those published by Bradbury.11 The dissimilarities may occur
because of differing climate, topography, or definition of the convective mixing-layer thickness.

                      Figure 5. Updraft mean vertical velocity ratio as function of height ratio.

                                                          7
                                  American Institute of Aeronautics and Astronautics
Equation (12) calculates updraft outer radius, r2, as determined by Lenschow.10

                                        !                 1               $
                                        #           ! z $3!       z$      &
                               r2 = max # 10, 0.102 # & # 1 ' 0.25 & ( zi &                                     (12)
                                        #"          " zi % "      zi %
                                                                          &%

   Equation (12) predicts small updraft outer radius near the ground and an increasingly larger updraft radius as
height increases. Figure 6 shows updraft outer radius ratio, r2/zi. Updraft radius is independent of velocity in this
model.

  Figure 6. Ratio of outer radius to convective mixing-layer thickness, r2/zi, as function of height ratio, z/zi.

    The distribution of vertical velocity within the updraft varies considerably between updrafts. The literature
contains various updraft vertical velocity distributions. Wharington7, 12 used Gaussian-shaped updrafts to study
autonomous soaring techniques and Metzger and Hedrick4 used simple square-edged lift regions to optimize flight
paths for cross-country flight. Many references depict updrafts to be quite random in size and updraft velocity
distribution. In the book Cross Country Soaring,13 Helmut Reichmann states that thermals are “seldom, if ever,
round” and “the thermals that depart from the norm are most likely the norm itself.”
    Flight test results were used in an attempt to model the distribution of vertical velocity for the updraft model
presented in this paper. Flight tests conducted by Konovalov14 show two basic vertical velocity profiles, called
type-a and type-b. Type-a updrafts are strong and have a wide area of nearly constant lift. Type-b updrafts are
weaker and have increasingly more lift toward the center. Konovalov also correlated the frequency of occurrence of
type-a and type-b updrafts to the updraft diameter. This paper uses Konovalov’s data to correlate the vertical
velocity distribution of an updraft to its outer radius. Figure 7 shows a revolved trapezoid that was used to fit
Konovalov’s type-a and type-b updrafts by adjusting the inner radius, r1. This study uses a revolved trapezoid
because the trapezoid provided a good approximation to the vertical velocity profiles given by Konovalov. A fit of
Konovalov’s data yields the piecewise function for the updraft radius ratio, r1 /r2, given in Eq. (13).

                                                        8
                                American Institute of Aeronautics and Astronautics
Figure 7. Revolved trapezoid vertical velocity distribution.

                                    r1   "0.0011! r2 + 0.14 for r2 < 600m
                                       = #                                                                          (13)
                                    r2   $0.8               else

   The average vertical velocity of a revolved trapezoid updraft distribution is found by dividing the volume of the
revolved trapezoid shape by the base area. Equation (14) gives the resulting relationship.

                                1
                                    2! % r1                      r2
                                                                      w peak ( r2 # r )          (
                         w=        $ ' $ w peak " r " dr + $
                              !r22 0 '& 0                                 r2 # r1
                                                                                          r " dr * d+
                                                                                                 *)
                                                                                                                    (14)
                                                           r1

   Evaluating Eq. (14) and solving for wpeak gives the relationship found in Eq. (15).

                                              w peak =
                                                             (
                                                         3w r23 ! r22 r1     )                                      (15)
                                                             r23 ! r13

    Equation (15) determines the peak value, wpeak, of a revolved trapezoid updraft distribution for a given average
updraft velocity, wT. The vertical velocity distribution was further refined by fitting a bell shape given in Eq. (16) to
approximate the revolved trapezoid distribution. A bell shape was chosen because the bell shape defines a smooth
vertical velocity distribution.

                                          "                               %
                                          $                               '
                                          $
                               w = w peak $
                                                  1
                                                             + k4 ! + wD ''
                                                                    r
                                                                                                                    (16)
                                                          k2       r2
                                          $1+ k ! r + k                   '
                                          $#   1
                                                  r2    3                 '&

   In Eq. (16), k1–4 are shape constants and wD is the downdraft velocity. This equation describes a family of curves
designed to fit the trapezoid shape given by Eq. (13) for a range of discrete r1/r2 values. Seven evenly spaced points

                                                         9
                                 American Institute of Aeronautics and Astronautics
were chosen to represent the range of possible vertical velocity distributions. Table 3 gives the shape constants used
in Eq. (16) for each r1/r2 ratio. Figure 8 shows the resulting family of updraft velocity distributions. The downdraft
velocity term, wD, simulates the toroid-like downward velocity found on the outer edge of updrafts as described by
Bradbury. Figure 9 illustrates the toroid shape of the updraft as the updraft increases in height. Eqs. (17) and (18)
calculate the downdraft velocity.

                   Table 3. Shape constants for bell-shaped vertical velocity distribution.
                            r1/r2               k1           k2           k3           k4
                            0.14              1.5352       2.5826      –0.0113       0.0008
                            0.25              1.5265       3.6054      –0.0176       0.0005
                            0.36              1.4866       4.8354      –0.0320       0.0001
                            0.47              1.2042       7.7904       0.0848       0.0001
                            0.58              0.8816       13.972       0.3404       0.0001
                            0.69              0.7067       23.994       0.5689       0.0002
                            0.80              0.6189       42.797       0.7157       0.0001

                               Figure 8. Bell-shaped vertical velocity distributions.

                                                            10
                                    American Institute of Aeronautics and Astronautics
Figure 9. Development of updraft toroid shape as height increases.

                                       * !" $ " # r '
                                       , sin &                    for r1 < r < 2r2
                                  w1 = + 6   % r2 )(                                                              (17)
                                       ,0                         else
                                       -

                                  (      " z     %                             z
                                  *2.5w1 $ ! 0.5 '                 for 0.5 <      < 0.9
                             wD = )      # zi    &                             zi                                 (18)
                                  *0                              else
                                  +

     Additionally, Eqs. (17) and (18) modify the vertical velocity shape given in Eq. (16) to have zero net vertical
velocity when z/zi = 0.9. The height ratio 0.9 was chosen because the vertical velocity of the updraft is negative at
altitudes above 0.9 as given in Eq. (11). Figure 10 gives vertical velocity profiles for low and high updrafts. The
vertical velocity profiles show increasing downdraft at the outer edge of the updraft as height is increased. Figure 10
was generated with the convective velocity scale, w*, and convective mixing-layer thickness, zi, of 2.56 m/s and
1401 m, respectively.

                                                         11
                                 American Institute of Aeronautics and Astronautics
Figure 10.    Vertical velocity profiles of a typical updraft with increasing height.

                              V.     Updraft Spacing and Environment Sink
    Updraft spacing was determined from Lenschow10 for a constant z/zi of 0.4. Equation (19) gives the resulting
relationship.

                                                    N L zi
                                                           = 1.2                                                 (19)
                                                     L
    In Eq. (19), NL is the number of updrafts encountered along a straight line of length, L. Equation (20) solves for
the number of updrafts within a given area of length, X, and width, Y, and a given updraft outer radius, r2.

                                                          0.6YX
                                                    N=                                                           (20)
                                                           zi r2

    In a given test area, all upward moving air in the form of updrafts is balanced by an equal amount of downward
moving air, known as environment sink, in the surrounding atmosphere. Environment sink is applied whenever the
aircraft is not inside an updraft. Conservation of mass was used to determine the environment sink velocity for a
given test area. Equation (21) gives the resulting relationship.

                                                              2
                                                      ! wN"r2 swd
                                               we =                                                              (21)
                                                      X * Y ! N"r22

   Equation (21) relates the environment sink velocity, we, to the average updraft velocity,          wT , using the
downdraft velocity ratio, swd, given in Eq. (22).

                                                            wD
                                                    swd =                                                        (22)
                                                            w1

                                                         12
                                 American Institute of Aeronautics and Astronautics
The downdraft velocity ratio is used to modify the environment sink calculation to account for the downward
moving air defined in Eqs. (17) and (18). The vertical velocity distribution of the updraft can be blended to match
the environment sink at the outer radius of the updraft using Eq. (23).

                                                    &    w          #
                                              wc = w$1 ' e          ! + we                                            (23)
                                                    $   w peak      !
                                                    %               "

    Equation (23) stretches the vertical velocity distribution to maintain the maximum vertical velocity at the center
of the updraft while allowing a smooth transition to the environment sink at the outer edge of the updraft.

                                                VI.     Application
    This paper presents an updraft model for a specific intention—to use the model during the design and simulation
of autonomous soaring UAVs. This section introduces an example set of updraft calculations for a typical simulation
case. Appendix A gives a MATLAB® (The MathWorks, Natick, Massachusetts) script for this example. The
convective scale factors, w* and zi, were taken from the mean values given in Table 1 to be 2.56 m/s and 1401 m,
respectively, for this example. The height, z, used in this example was chosen as 280 m, giving a height ratio, z/zi, of
0.2. Calculating the updraft outer radius, r2, from Eq. (12) yielded 79.4 m. The number of updrafts in the test area
(defined by X = 1000 m and Y = 1000 m) calculated from Eq. (20) was five, using r2 = 79.4 m and zi = 1401 m. The
resulting five updrafts can be positioned anywhere within the test area. Updrafts were positioned evenly along a
diagonal line in this example for clarity. The vertical velocity of the test area was calculated with Eqs. (11) to (18)
and (21) to (23) of this paper for every point within the test area using a grid spacing of 10 m. Figure 11 shows the
resulting vertical velocities of the test area. Figure 12 shows the vertical velocity distribution of a single updraft as a
check case for the code given in Appendices A and B. Figure 12 of this paper was produced with the MATLAB
script given in Appendix A. Electronic copies of this code can be obtained from the author.

Figure 11. Example test area (calculated from MATLAB code given in Appendix A). Updrafts are
positioned along a line for clarity.

                                                          13
                                  American Institute of Aeronautics and Astronautics
Figure 12.    Vertical velocity profile check case (calculated from MATLAB code given in Appendix A).

    The updraft positions should be randomly chosen during an actual application. Updraft positions should be held
for 20 min of simulation time and then re-calculated to account for the duration of a typical updraft. Updraft
duration can range from 5 to 30 min, indicating that this parameter should be varied during simulation testing of
autonomous soaring algorithms.

                                            VII.     Future Work
   The updraft model presented in this paper is useful for the development of autonomous soaring guidance and
control but does not model all of the significant characteristics of naturally occurring updrafts. Naturally occurring
updrafts have time dependant vertical velocities and radiuses. Additionally, naturally occurring updrafts drift and
change shape when wind is present; and naturally occurring updrafts merge together to form larger and stronger
updrafts as height increases. Downdrafts are also known to form along with updrafts during times of strong
convection. Future updraft models, therefore, should include these effects as well as include convective scale
parameters for locations and topographies other than the desert.

                                       VIII.    Concluding Remarks
    This paper presented a model of convective updrafts using convective scale parameters calculated from ground
and balloon measurements taken at Desert Rock, Nevada. Convective scale velocity and convective mixing layer
thickness test cases were produced using statistical tools. Statistical representation did not include measurement data
during times of zero convection, representing 25 percent of the total measurement set. A statistical representation of
the convective mixing-layer thickness values corresponding to the set of convective scale velocities was produced
by fitting the data to a Gamma probability density function. Convective scale parameters were then used to calculate
mean updraft vertical velocity and radius using equations from Lenschow. The vertical velocity distribution within
the updraft was calculated by fitting the data in Konovalov to a family of bell-shaped curves. The toroid structure of
updrafts as height increases, described in Bradbury, was included in the vertical velocity distribution of this model.
Updraft spacing was calculated from a relationship given in Lenschow; and environment sink was calculated using
the conservation of mass. This paper also presented an example use of the equations within the paper using the
MATLAB code provided in the appendices. Taken together, the equations given herein describe an updraft model
that is suitable for the preliminary design and simulation of guidance and control for soaring uninhabited air
vehicles.

                                                         14
                                 American Institute of Aeronautics and Astronautics
Appendix A.    MATLAB Code for Example Application
%This script will run a check case of the NASA DFRC updraft model.
%
%Michael J. Allen
%NASA Dryden Flight Research Center
%2005
%

%DEFINE UPDRAFT PARAMETERS
wstar=2.56;                                        %selected w*, m/s
zi=1401;                                           %selected zi, m
z=280;                                             %test height, m
wgain=1;                                           %multiplier on vertical velocity
rgain=1;                                           %multiplier on radius

%DEFINE AREA
X=1000;                                            %length of test area, m
Y=1000;                                            %width of test area, m
A=X*Y;                                             %test area, m^2

%CALCULATE OUTER RADIUS
zzi=z/zi;
r2=(.102*zzi^(1/3))*(1-(.25*zzi))*zi;              %Equation 12 from updraft paper

%CALCULATE NUMBER OF UPDRAFTS IN GIVEN AREA
N=round(.6*Y*X/(zi*r2));

%SET PERTURBATION GAINS FOR EACH UPDRAFT
wgain(1:N)=1;                            %multiplier on vertical velocity
rgain(1:N)=1;                            %multiplier on radius

%PLACE UPDRAFTS IN A LINE (NORMALLY THIS IS DONE RANDOMLY)
for kn=1:N,         %for each updraft
    xt(kn)=kn*X/(N+1);
    yt(kn)=kn*Y/(N+1);
end

%DEFINE GRID OF TEST LOCATIONS
xc=[0:10:X];
yc=[0:10:Y];
xx=xc'*ones(1,length(yc));              %create matrix of x values
yy=ones(length(xc),1)*yc;               %create matrix of y values
zz=ones(size(xx)).*z;                   %create matrix of z values
for kx=1:length(xc),
    disp([num2str(kx) 'of ' num2str(length(xx))])
    for ky=1:length(yc),

            %CALL UPDRAFT FUNCTION
            w(kx,ky)=run_model2_3(xx(kx,ky),yy(kx,ky),zz(kx,ky),…
            xt,yt,wstar,wgain,rgain,zi,A,1);
      end
end

%PLOT UPDRAFT FIELD.      THIS WILL CREATE FIGURE 11
figure; orient tall
  mesh(xx,yy,w)

                                                 15
                         American Institute of Aeronautics and Astronautics
xlabel('X position, m')
  ylabel('Y position, m')
  zlabel('w, m/s')
  set(gca,'View',[-22 64])

%PLOT CROSS-SECION OF 1ST UPDRAFT. THIS WILL CREATE FIGURE 12
figure; orient tall
  h1=plot(yy(18,1:40),w(18,1:40),'k-'); grid
  set(gca,'fontsize',12)
  set(h1,'linewidth',2)
  xlabel('y position, m')
  ylabel('w, m/s')

                   Appendix B.    MATLAB Code for Updraft Model
function [w,r2,wc] = ...
run_model2_3(x,y,z,xt,yt,wstar,wgain,rgain,zi,A,sflag)
  %function [w,r2,wc] = run_model2_3(x,y,z,xt,yt,wstar,wgain,rgain,zi,A)
%
%Input: x       = Aircraft x position (m)
%        y      = Aircraft y position (m)
%        z      = Aircraft height above ground (m)
%        xt     = Vector of updraft x positions (m)
%        yt     = Vector of updraft y positions (m)
%        wstar = updraft strength scale factor,(m/s)
%        wgain = Vector of perturbations from wstar (multiplier)
%        rgain = Vector of updraft radius perturbations from average
%                 (multiplier)
%        zi     = updraft height (m)
%        A      = Area of test space
%        sflag = 0=no sink outside of thermals, 1=sink
%
%Output: w = updraft vertical velocity (m/s)
%        r2 = outer updraft radius, m
%        wc = updraft velocity at center of thermal, m/s
%
%
%Michael J. Allen, NASA DFRC, 2005

%DEFINE UPDRAFT SHAPE FACTORS
r1r2shape = [0.1400 0.2500 0.3600     0.4700 0.5800 0.6900 0.8000]';
Kshape = ...
    [1.5352    2.5826   -0.0113       -0.1950         0.0008;...
     1.5265    3.6054   -0.0176       -0.1265         0.0005;...
     1.4866    4.8356   -0.0320       -0.0818         0.0001;...
     1.2042    7.7904    0.0848       -0.0445         0.0001;...
     0.8816   13.9720    0.3404       -0.0216         0.0001;...
     0.7067   23.9940    0.5689       -0.0099         0.0002;...
     0.6189   42.7965    0.7157       -0.0033         0.0001];

%CALCULATE DISTANCE TO EACH UPDRAFT
N=length(xt);
for k=1:N,
    xdsq=(x-xt(k))^2;
    ydsq=(y-yt(k))^2;
    dist(k)=sqrt(xdsq+ydsq);
end

                                             16
                     American Institute of Aeronautics and Astronautics
%CALCULATE AVERAGE UPDRAFT SIZE
zzi=z/zi;
rbar=(.102*zzi^(1/3))*(1-(.25*zzi))*zi;

%CALCULATE AVERAGE UPDRAFT STRENGTH
wtbar=(zzi^(1/3))*(1-1.1*zzi)*wstar;

%USE NEAREST UPDRAFT
upused=find(dist==min(dist));
if length(upused)>1; upused=upused(1); end

%CALCULATE INNER AND OUTER RADIUS OF ROTATED TRAPEZOID UPDRAFT
r2=rbar*rgain(upused); %multiply by random perturbation gain
if r2
kd=Kshape(5,4);
        elseif r1r2A, error('Area of test space is too small'); return; end

if sflag,
    we=-(At*wtbar*(1-swd))/(A-At);                    %environment sink rate,
                                                      %positive up (m/s)
    we(find(we>0))=0;                                 %don't allow positive sink
else,
    we=0;
end

%STRETCH UPDRAFT TO BLEND WITH SINK AT EDGE
if dist(upused)>r1;         %if you are outside the core
    w=w2*(1-we/wc)+we;      %stretch
else,
    w=w2;
end

                                                18
                        American Institute of Aeronautics and Astronautics
References
   1
       “Fédération Aéronautique Internationale (FAI) – Gliding World Records,” Fédération Aéronautique Internationale, URL:
http://records.fai.org/gliding/current.asp?id1=DO&id2=1 [cited 08 November 2005].
   2
      Weimerskirch, Henri, Chastel, Olivier, Barbraud, Christophe, and Tostain, Olivier, “Frigatebirds Ride High on Thermals,”
Nature, Vol. 421, January 23, 2003, pp. 333–334.
    3
      Allen, Michael J. “Autonomous Soaring for Improved Endurance of a Small Uninhabited Air Vehicle,” AIAA-2005-1025,
43rd AIAA Aerospace Sciences Meeting, Reno, NV, January 10–13, 2005.
    4
      Metzger, Darryl E., and Hedrick, J. Karl, “Optimal Flight Paths for Soaring Flight,” AIAA 74-1001, AIAA/MIT/SSA
 nd
2 International Symposium on the Technology and Science of Low Speed and Motorless Flight, Cambridge, Massachusetts,
September 11–13, 1974.
    5
      Cone, Clarence D., Jr., “The Design of Sailplanes for Optimum Thermal Soaring Performance,” NASA TN D-2052, 1964.
    6
      Mostovoi, Gueorgui V., “Two-Dimensional Lagrangian Model of a Rising Thermal: Application for a Liquid Water Content
Spatial Variability of a Warm Cumulus Cloud,” Atmospheric Research 43, 1997, pp. 233–252.
    7
      Wharington, John, and Herszberg, Israel, “Control of a High Endurance Unmanned Air Vehicle,” ICAS-98-3,7,1, AIAA
A98-31555, 21st ICAS Congress, Melbourne, Australia, September 13–18, 1998.
    8
      “The       SURFRAD         Network,”       National     Oceanic    and        Atmospheric       Administration,   URL:
http://www.srrb.noaa.gov/surfrad/index.html [cited 27 September 2004].
    9
      Stull, Roland B., An Introduction to Boundary Layer Meteorology, Kluwer Academic Publishers, Norwell, MA, ISBN 90-
277-2769-4, 1994.
    10
       Lenschow, D. H., and Stephens, P. L., “The Role of Thermals in the Convective Boundary Layer,” Boundary-Layer
Meteorology, 19, 1980, pp. 509–532.
    11
       Bradbury, Tom, Meteorology and Flight, A Pilot’s Guide to Weather, 3 rd Edition, A & C Black Ltd, London, Great Britain,
ISBN 0-7136-4226-2, 2000.
    12
       Wharington, John, and Herszberg, Israel, “Optimal Semi-Dynamic Soaring,” Royal Melbourne Institute of Technology,
Melbourne, Australia, October 11, 1998.
    13
       Reichmann, Helmut, Cross Country Soaring, Soaring Society of America, Inc., Hobbs, NM, ISBN 1-883813-01-8, 1993.
    14
       Konovalov, D. A., “On the Structure of Thermals,” 12th OSTIV Congress, Alpine, USA, 1970.

                                                            19
                                    American Institute of Aeronautics and Astronautics
You can also read